Font Size: a A A

Research On FMCW Millimeter Wave Radar Target Feature Extraction And Classification

Posted on:2021-03-25Degree:MasterType:Thesis
Country:ChinaCandidate:C ChenFull Text:PDF
GTID:2428330611955169Subject:Engineering
Abstract/Summary:PDF Full Text Request
The frequency-modulated continuous wave signal transmitted by the millimeterwave radar is mixed with the signal received by the radar to extract and classify the target.This thesis studies the target feature extraction and classification method of FMCW millimeter wave radar.The specific content includes the following parts:1.Introduce the system composition of millimeter wave radar,mainly understand the triangle wave signal waveform and study the signal analysis theory of frequency modulation continuous wave,and deduce the principles of range measurement speed and angle measurement of millimeter wave radar.2.Researched a high-precision angle measurement algorithm for DOA estimation,analyzed the array receiving model of uniform linear array,DOA estimation algorithm,target lateral characteristics and high-precision angle extraction method.The simulation verifies that the high-precision angle measurement method has a very high angular resolution,and the experiment proves that the high-precision angle measurement information can be used as a method for classifying targets.3.A millimeter-wave radar detection algorithm based on Doppler extended targets is studied,Chirp waveform signals and Doppler extended characteristics are analyzed,and Doppler extended target detection procedures and detection processes are analyzed.Simulation verification of Doppler extended target detection method proves that the method of accumulating Doppler frequency band has superior detection performance.4.Researched three micro-Doppler feature extraction algorithms,established pedestrian motion models,studied the principles of short-time Fourier transform,wavelet transform,and Wigner-Weir distribution,and used three methods for pedestrian motion models Carry out simulation analysis and compare the time-frequency analysis results of the three methods,proving that all three methods have time-frequency analysis capabilities.5.Perform classification test of micro-Doppler features on the measured data,introduce the platform and parameter configuration of the measured data collection and the field test environment,perform micro-Doppler feature extraction and classification on the measured data after micro-Doppler processing,The experimental results are analyzed to verify that the micro-Doppler feature can distinguish pedestrian vehicles and trees with high accuracy.The micro-Doppler feature extraction and classification method is a good target classification method.
Keywords/Search Tags:target feature extraction and classification, DOA estimation algorithm, Doppler extended target detection algorithm, short-time Fourier transform
PDF Full Text Request
Related items